In recent years, the U.S. Marine Corps has begun developing an infrastructure for applying agent-based models and simulation, computing power, and data analysis and visualization technologies to help answer complex questions in military operations. Factor screening approaches are of particular interest, since even relatively simple agent-based models may have hundreds (or even thousands) of inputs that can be varied. We describe a new experimental design, called a frequency-based design, that can be used for exploring the behavior of terminating simulations. We apply this to a model of a peace-enforcement operation. We examine the behavior of four performance measures (including two attrition ratios) and discuss how the results confirm and complement earlier findings. We conclude with a brief discussion of issues that merit further investigation.